Search

A general definition and corresponding measurement rules are provided for voltage resolution of contactless chip testing systems. The emphasis is on a practical measurement rule which allows the determination of the voltage resolution solely from parameters which are accessible to every user. To achieve this goal, first two characteristic parameters are determined which define voltage resolution from the physical effects limiting the resolution and then a measurement rule is provided to measure these parameters by means of a well defined test signal. This procedure will allow an unambiguous comparison of contactless testing systems as well as of the results obtained with them.

A voltage detector and its connected focusing system for voltage measurements on integrated circuits by laser induced photoemission are presented. The capabilities of the system are analyzed theoretically and verified experimentally. Application to metal lines of submicrometer width and spacing is possible, combined with picosecond time resolution and a voltage sensitivity far exceeding that of electron beam probing. An excellent insensitivity to static crosstalk perturbations is demonstrated. Voltage contrast measurements and measurements of photoemission spectra have also been performed with this system.

High-quality etched mirrors for AlGaAs/GaAs power lasers for applications in optical storage have been fabricated by chemically assisted ion-beam etching (CAIBE). In order to ensure flat mirror facets of the ridge-waveguide lasers, a flared-waveguide end section is used. This results in a very slight mirror roughness of approximately 20 nm across the beam cross section, and yields excellent beam properties allowing diffraction-limited focusing up to 50-mW output power. >

The rapid evolution of advanced VLSI and VHSIC technologies, with advanced devices switching on time scales of less than 10psec creates new demands on high-speed diagnostic and testing instruments. The most important demands are the capability of contactless and noninvasive measurement of switching times and voltage states on internal nodes of VLSI ICs with a high time resolution (1–5 psec), a high voltage resolution (1–10 mV) in conjunction with a short measurement time, and a spatial resolution in the submicron range. It should be emphasized that these requirements have to be achieved simultaneously. In addition, it is desirable to have a diagnostic technique that is applicable to all semiconductor technologies rather than being limited to exclusively silicon- (Si) or gallium-arsenide- (GaAs) based circuitry.

There has been a strong effort recently to develop so-called vectorial beam propagation methods (VBPM) [1-7]. The aim of this effort is to overcome the scalar limitation of the classical beam propagation method (BPM) [8-10]. We present here a new approach to the VBPM that can handle three-dimensional anisotropic media. This new method allows the study of the vectorial properties of the electromagnetic fields, such as polarization effects and coupling between the field components.

A finite-difference beam-propagation method is derived directly from the vectorial Maxwell equations. The new approach allows the analysis of coupling between various components of the electric-field vector for optical beams propagating in dielectric waveguides as well as corresponding changes in the beam-polarisation state.

It is shown that the response of the electrostatic potential on voltage changes at a conducting line in a microscopic structure is fully described by a pure geometry function. This function is determined by the special microscopic character of the conducting line and by the screening due to neighboring conductors. It already allows a qualitative description of transit‐time and cross‐talk effects in voltage measurements via electron emission or electro‐optical sampling, and provides a basis for the quantitative calculation of these effects. The geometry dependence of this function will be analyzed in detail.

A spin polarized photoemission experiment using transverse magnetized samples is described in detail. We discuss the advantages of this set-up in comparison to the one used so far which employed the longitudinal magnetization (polarization) geometry. Data on Ni(110) are presented including oxygen adsorption results, and discussed along current ideas in spin polarized photoemission theory.

The combination of high energy and angle resolution (100 meV, \ifmmode\pm\else\textpm\fi{}3\ifmmode^\circ\else\textdegree\fi{}) with spin analysis in photoemission spectroscopy is shown to allow the direct observation of magnetically spin-split electronic states. This is demonstrated for normal emission from Ni(110) with polarized Ne i radiation ($h\ensuremath{\nu}=16.85$ eV). It is found that ${\ensuremath{\Delta}}_{\mathrm{ex}}=0.18$ eV for the exchange splitting of the ${S}_{4}$ band at the $X$ point in good agreement with a band-structure and photoemission calculation that considers the self-energy correction explicitly.

High electron density effects on photoyield and electron energy distributions have become important in several electron spectroscopies and in electron microscopy since the availability of high‐intensity photon and electron sources. In this work we show that such effects also lead to limitations in voltage sensitivity and time resolution in photoemission and electron‐beam sampling measurements of voltage signals if picosecond time resolution is required. The influence of high electron densities in picosecond pulses for these methods is analyzed theoretically and compared with experimental results in photoemission sampling. It is shown that the space‐charge limitation to the number of electrons per pulse which can be emitted from a sample and the energy broadening of the pulse owing to the Coulomb repulsion between the electrons impose significant limitations on the voltage sensitivity of photoemission sampling and of electron‐beam sampling with a pulsed high‐intensity cathode. Also the time resolution is d...

In this article, the transit‐time and cross‐talk effects occurring in voltage measurement via electron emission are analyzed as a function of geometry, extraction fields, electron start energies, and rise times of the actual input signals. The investigation was performed to obtain information about the maximum achievable time resolution and the disturbance of measured signals by cross‐talk effects in corresponding contactless measurement techniques like photoemissive sampling and electron‐beam probing. The field distribution above the sample was assumed as being two‐dimensional, and the influence of space‐charge effects neglected. A strong correspondence to the purely geometry‐dependent response function of the potential energy, analyzed in Part I [J. Appl. Phys. 62, 1553 (1987)], is established.

Recent achievements in solving the problem of the microscopic nature of the ferromagnetic-to paramagnetic-phase transition of ferromagnetic metals by employing the method of spin- and angle-resolved photoemission spectroscopy with synchrotron radiation (from the storage ring BESSY in Berlin) are briefly summarized. Emphasize is also given to the theoretical interpretation of the data, from which a short-range order in Fe near the Curie pointTc of at least 4 A has been derived. This excludes the validity of the disordered local moment picture for Fe at high temperatures.

A novel approach to contactless measurement of voltages on internal nodes of integrated circuits is presented. The method is based on time-resolved photoemission exploiting the single-photon process with short laser pulses in the ultra-violet. The method allows spatial resolution in the submicron range given by the diffraction limit for UV photons, a time resolution of a few picoseconds given by the width of the laser pulses and the electron transit-time effect, and a voltage resolution of a few microvolts within a signal integration time of one second.

Using spin-polarized photoemission with high energy- and angle resolution (ΔE=100 meV, ΔΘ=±3°) we have investigated the temperature dependence of the exchange splitting of Ni in the temperature range 0.5≦T/T c≦0.94. At room temperature we findΔ ex=0.18 eV for the exchange splitting of theS 4 band at theX point of the Brillouin zone. With increasing temperature the total (spin-averaged) energy distribution shows a narrowing and merges into one peak. The spin-resolved energy distribution curves approach each other and are strongly broadened. A discussion of the data within current theories of itinerant electron magnetism is given. The spectra indicate that neither the pure Stoner model nor the prediction of local band theory, assuming a temperature independent exchange splitting are justified for Ni. We conclude that the exchange splitting decreases with increasing temperature and that transverse as well as longitudinal spin fluctuations are responsible for the broadening of the spin-resoived energy distribution curves.

Calculations of spin-polarized threshold-photoemission spectra for Ni(001), Ni(111), and Ni(110) surfaces are presented, obtained with a one-step model of the photoemission process. As in the previous work of Moore and Pendry, good agreement with the angle-integrated experimental results for emission from the (001) and (111) surfaces is obtained within a normal-emission approximation. In contrast, for emission from the (110) surface it is found that the angle-integrated spectra are dominated by non-normal-emission contributions. In addition, the influences of relativistic dipole selection rules and surface-state emission on the spin polarization from the (110) surface are discussed briefly.

An electron spectrometer allowing energy‐ and angle‐resolved spin‐polarized photoemission from ferromagnets is described. The system is designed for use with samples magnetized perpendicular to the photoelectron extraction direction (transverse magnetization geometry) and a Mott detector (100 keV) for the spin‐polarization measurements. It transforms the emitting spot at the magnetic photocathode to a parallel beam which matches the size of the apertures of the Mott detector, resulting in a high efficiency. Angle selection is achieved despite initial acceleration of the photoelectrons. Energy resolution is obtained with a 90° spherical deflector connected to the Mott detector whose apertures act as the spectrometer exit slits. The performance of the setup is demonstrated by measurements with an evaporated iron film and a neon resonance lamp as light source. Sample geometries for use with the transverse magnetization configuration are discussed.

After a short review of the basic principles of time-resolved photoemission and its capabilities as a method for contactless electrical testing of integrated circuits, this paper reports on the progress made toward meeting the requirements for integrated-circuit (IC) testing and diagnostic equipment. The goals of IC testing are to provide high time and voltage resolutions with short testing time, and simultaneously high spatial resolution independent of the circuit technology (Si, GaAs). Here, the emphasis is on the time and voltage resolutions and short testing time. In the photoemission process, these parameters are independent of the spatial resolution in the regime of interest. The investigation is carried out with a pulsed-laser system generating picosecond pulses in the UV range. The measurements presented were performed on photoconductive switches. This allows testing the time resolution of the method by ultra-fast signals, and generating the electric transients to be sampled by the same laser source as the sampling pulse, guaranteeing almost jitter-free operation. The results clearly demonstrate that photoemission sampling is capable of contactless diagnostics and testing of very high-speed circuits with a high-voltage resolution and short testing time.

The dependenc of spin-resolved photoemission and inverse photoemission spectra on short-range magnetic order below and aboveT c is investigated within a bulk interband transition model. The spectra are calculated in a 2000-atom cluster of bcc-Fe with local exchange field configurations, which are random subject to an assumed spin-correlation function. The model continuously bridges the gap between the “disordered-local-moment picture” (no short-range order) and a “local-band picture” (massive short-range order). Special emphasis is given to the analysis of factors like group velocity and the symmetry of the wave function by investigating special points in the Brillouin zone. Comparison with experimental data reveals a short-range order of at least 4 A (corresponding to a nearest-neighbour correlation of about 0.4) nearT c .

The total density of states has been calculated for 2000-atom cubic Fe clusters with local exchange fields, which vary randomly subject to assumed degrees of magnetic order. Comparison with X-ray photoemission and appearance potential data near the Curie temperature rules out a disordered-local-moment picture and suggests short-range ferromagnetic order extending over at least 4 A up to about 6 or 7 A.

The newly developed method of photoemission sampling, based on the single‐photon process, has been applied to measure the dispersion of a picosecond voltage pulse traveling on a transmission line. The dispersion of the propagating pulse is clearly revealed in the spectra which show signal rise times of about 17 ps at 2 mm away from the pulse generating switch, and change into rise times of about 30 ps at 1.5 mm further away. The time resolution of the measurements is better than 7 ps. Therefore, photoemission sampling with picosecond laser pulses permits the measurement of changes in signal waveforms at arbitrary points of a transmission line and for arbitrary substrate materials with sufficient time resolution.

It is shown by electron-energy-loss spectroscopy with spin analysis that electrons inelastically scattered from a ferromagnet (${\mathrm{Fe}}_{82}$${\mathrm{B}}_{12}$${\mathrm{Si}}_{6}$) can have a high spin polarization as a result of exchange scattering. A maximum of the spin polarization occurs around 2.2-eV energy loss corresponding to the ferromagnetic exchange splitting. This experiment offers new possibilities for studying spin-flip Stoner excitations in ferromagnets.

The dependence on geometry, extraction field, electron start energy, and rise time of the input signal is investigated for the transit-time and cross-talk effects in stroboscopic voltage measurements via electron emission. The investigation yields information about the best achievable time resolution and the disturbance of measured signals by signals on neighboring conductors in corresponding contactless voltage-measuring techniques like photoemission sampling and electron-beam sampling.

The growing complexity of integrated circuits (IC) and the rapid development in VHSICs require contactless testing methods for logic state analysis and waveform measurements on internal nodes of VLSI chips. Considering the fast evolution of VLSI/VHSIC technologies, a contactless testing method has simultaneously to provide challenging features like high-voltage resolution (mV), high spatial resolution (sub-micrometer), and high time resolution (ps) in a short testing time. In addition to these requirements, it is desired that the method should be applicable to all technologies, e.g., Si as well as GaAs-based ones.

The idea of utilizing time -resolved photoemission for contactless electrical testing on internal nodes of IC's israther young. Initial studies quickly showed the advantages the technique offers over other methods. After a shortreview of the basic principle and its attractive features, this paper reports on the progress made toward meeting therequirements for IC testing and diagnostic equipment. The goals of IC testing are to provide high time and voltageresolutions with short testing time, and simultaneously high spatial resolution independent of the circuit technology(Si, GaAs). Here, the emphasis is on the time and voltage resolutions and short testing time. In the photoemissionprocess, these parameters are independent of the spatial resolution in the regime of interest. The investigation iscarried out with a pulsed -laser system generating picosecond pulses in the UV range. The measurements presentedwere performed on a photoconductive switch. This allows testing the time resolution of the method by ultra -fastsignals and generating the electric transients to be sampled by the same laser source as the sampling pulse. Thelatter point guarantees almost jitter -free operation. The results clearly demonstrate that photoemission sampling iscapable of contactless diagnostics and testing of very high -speed circuits with a high voltage resolution and shorttesting time.IntroductionThe major challenge for contactless IC testing is the combination of the three measuring parameters: 1) Timeresolution in the low picosecond range; 2) spatial resolution in the submicrometer range, and 3) voltage resolutionbelow 10 mV accomplished with short testing time as well as a wide range of applicability.Only a few years ago, was there just one feasible method to probe internal nodes of integrated circuits bycontactless means: electron beam probing.1-3 The ever -increasing switching speed and decreasing voltage swingsof modern high -speed electronic circuits have exceeded the performance limits of existing e-beam testing

The basic principles, fundamental limits, and recent results of the novel photoemission probing method are reviewed. The possible range of applications as a contactless diagnostic method for integrated circuits in the real-time and sampling modes of photoemission probing is discussed on the basis of theoretical and experimental studies. In photoemission sampling, electron emission is induced in a sample by an incident light beam (in most cases a laser beam), and the photoemitted electrons are then accelerated towards a reference or retarding field electrode. If the reference electrode is at the same potential as the point of electron emission on the VLSI circuit, all photoemitted electrons can pass the reference electrode and are guided towards an electron detector. >

After giving brief background information on photoemission testing with emphasis on measurement time, the authors describe progress made in detector design and viewing capability. The usefulness of equipment, using the photoemission sampling technique has been demonstrated through measurements delay and rise times of integrated GaAs MESFET circuits. The switching time of these circuits is around 200 ps, therefore presenting no challenge to the time resolution of the system, which had been substantiated by earlier measurements of electrical pulses with 8-ps rise time. The ability to activate devices within a chip by a pulsed visible light beam is also demonstrated. >